I am agree.
I am agree.[/QUOTE]
If anybody cares, I have been doing exactly that for a while.
http://kokcito.tk/mhc (female nudity in images)
Creating targets for your own model is a great way to create lots of variations, without always using the error prone shape keys. The base mesh is not hardcoded in any flavor of Makehuman and the functions to import a new mesh to Makehuman 1.8 and model new targets are already there, even if there is not a “create new base” button.
I know your site and the mesh of the female you share and your tutorials (that are really good). The images in the link that you posted in this post are really impressive, but it seams to me that your character is a little too rigid in some poses. Probably a problem of rigging more than a problem of modeling.
Anyway it would be nice to discuss more about modeling details with someone like you that have a lot of experience experimenting with realistic human models and have reached such high results.
Thanks Ender. I hope to start a WIP or focused critique thread sometime when I have better results…
Wow… thank you for this… I really havent had the time or knowledge to even begin working on good human-type models, but I think with your post and links I can finally start on it.
Any chance of getting the images in the first post back up? I’d love to see what they were!
I have udated them. But I remind you that the notes are old and I have learned new thing that make me change many things in that modelling approach.
Muscles sketches I done are based on an erroneous reference book. Pectorals are wrong. But I drew them that way because someone critiqued the way pectorals are attached to the shoulder/arm in my model.
I was not sure of that but looking at my pectorals and those of other men bodyes it seamed that I was working in the right direction. Infact looking at some more precise reference books I discovered that the shape of my model was right.
But now I still need to understand how to organize mesh polygons in the upper pectorals. I am not sure that the polygons strands I used are positioned in the best way.
Moreover I think that I have used not enough subdivision of the starting poligons. Good one probably is 16 for torso and 12 for arms.
Thanks.
The pecs do attach the shoulder at several places. If they didn’t the pecs would be useless. They pull the arms inward and the attachement point is key to this. So one part of the pec would have attach to the humerous. Now I have seen topology where the pecs attach to the outside of the deltoid. This may be ideal for animation purposes. I think you could model the human with less polygons than that. That’s going to be a rather heavy mesh.
womball: this is the correct pectorals muscles structure. As you can see the full muscle torque as showed by green arrows. Clavicle part of the muscle goes above the breast part. This makes appearing it - outside (skin) - as if the muscle follows the line described by blue arrow. Exactly the shape I tryed to emulate in my model.
This is what I mean! Although originally I didn’t see it encorporated in the model. Carry on! 
Maybe I am blind, but the final mesh of the model I made actually follow that shape (blue arrow).
Perhaps the final does, I haven’t seen that, but the original drawings and model you had did not.
Please, there is anybody that can tell me if I am mad and I see ghosts in my model or if is womball to be wrong?
Please! Help!
I tryed a new approach. This time I followed muscles more preciselly with polygon strands. Even if I don’t know how much usefull can be this effort in animation.
Anyway the mesh is not completed, I am studing how to connect arm now, to the new strands.
Some basic explanations.
Violet area: I deleted all the polygons in the violet area sobstituting them with the new pectorals.
Yellow area: I created the long strand rapresented by the yellow area. It is the part of the pectorals attached to clavicle. Part of the polygons of this strand already exist in the deltoid mesh from the original model (look at main post).
Red area: I obtained the new pectorals extruding edges from lower part of the chest. Than i connected them with clavicle pectoral and subdivided them (CTRL+R) 2 times.
Blue area: Blue area have been addedd to add detail on shoulder.
Here the result without area hilight.
Two of the best tutorials I found, for example, the first for High Poly and the second for Low Poly modeling, are Modeling Joan of Arc by Michel Roger
I can’t understand why everybody finds the Joan of Arc tutorial the best tutorial out there. I find it to be the worst tutorial out there, and sending beginners over there to learn about organic modeling is the worst thing you can do. After every step you are left with “WHY?” Have you ever seen a beginner that became a master modeler by learning spoonfed tutorials?
I struggled a lot over the years with organic modeling and I consider myself an intermediate now. I suggest you to go over at www.subdivisionmodeling.com
Go to the sharing techinques forum and look for Form and Poles thread. The poster SomeArtist realy does a good job in explaining form and topology. A added some little posts too.
I think poly by poly technique is not the way to go, although most profesionals use it and Blender is excellent for poly by poly, but my objection is that you don’t have a rough overview of your topology. You are building lots of localized topologies that may clash when they are stich together. I find the concepts like Key loops and Fill loops that Some Artist talks about very clear to understand and execute
Two of the best tutorials I found, for example, the first for High Poly and the second for Low Poly modeling, are Modeling Joan of Arc by Michel Roger
I can’t understand why everybody finds the Joan of Arc tutorial the best tutorial out there. I find it to be the worst tutorial out there, and sending beginners over there to learn about organic modeling is the worst thing you can do. After every step you are left with “WHY?” Have you ever seen a beginner that became a master modeler by learning spoonfed tutorials?
I struggled a lot over the years with organic modeling and I consider myself an intermediate now. I suggest you to go over at www.subdivisionmodeling.com
Go to the sharing techinques forum and look for Form and Poles thread. The poster SomeArtist realy does a good job in explaining form and topology. A added some little posts too.
I think poly by poly technique is not the way to go, although most profesionals use it and Blender is excellent for poly by poly, but my objection is that you don’t have a rough overview of your topology. You are building lots of localized topologies that may clash when they are stich together. I find the concepts like Key loops and Fill loops that Some Artist talks about very clear to understand and execute
Yes! Absence of clear explanations is the same reason I don’t like many tutorials out there. And I am trying to cover this lack with my thread (that I remind is a note/diary of what I learn and the experiments I do). But considering the fact that online there is not much more it is one of the best tutorials you can find.
Yes I know it. It is a great resource. And thanks for the guide lines to find informations in that site. I made many searches there but never found what I need. I probably used wrong keywords and looked in wrong places.
Absolutelly agree with you. I think that starting from some primitive structure that roughly draw body shape is a great help to reach the correct result.
But the reason that leads me to make some patched model is not for the final result but for understanding how to subdivide logically the model of a body. In the beginning of the thread I tryed to color the different regions to best understand and synthetize in our mind the basic units/bricks of the body model. Once reached that knowledge it would be easier to create the rough mesh from wich starting all the models I want.
I choosed the arm/chest connection because it is the most problematic area of the body mesh. I have tryed many versions but I always obtained bad aesthetic results, even if almost correct from the point of view of anatomy.
And this is a strange thing appening with modeling, probably the reason is that in some cases and in some body regions it is better to have more subdivisions.
But at this point, the question I want to give an answer is: which is the best base mesh to start from? Considering this prerequisites: 1) it needs to be not highpoly and easly undestandable at a glance; 2) it needs to roughly but preciselly draw the whole body; 3) it needs to be the best base mesh from wich obtain the final highpoly model.
Hi all,
This has been very helpful for me-- I’m in the middle of making a simple character for demo reel work and halfway into weight painting I realized I’d done the shoulders all wrong.
I’m still having trouble, however. My topology isn’t perfect, but I can deal with that; the main issue I’m coming up against is automating rotations.
This is hard to explain in words, so I’ll show it in pictures. Bear in mind this is neither the final topology nor the final set of weights; I just rebuilt the shoulder so the the weights are a mess:
Rested pose. This is a bad attempt at your shoulder geometry, which I’m going to go back and rebuild some more later today.
Bicep bone rotated along global Y axis, close to 90 degrees, aiming the arm down. The deformation here is alright; what I don’t like about it I think I’ll be able to fix with proper weighting. However:
If you rotate down, to a true “rest pose” (none of us walk around with our arms out in the T-pose), then rotate forwards, you get nastiness. Now, the obvious fix for this is to adjust the rotation of the bicep:
In this image I’ve rotated the shoulder bone forwards as well as twisted the bicep on its lengthwise axis. Suddenly everything flattens out in a more pleasing manner.
Now the question: is it possible to get this to happen automatically? I like to automate little things like that as much as possible when rigging. I want to get it to a point where I can move the arm, rotating the bicep on its pitch and bank, but not along its lengthwise axis. Am I making sense? Is this possible, or do I have to always be mindful of all three axes when animating?
Tolobán, you seem to have this issue solved quite well in the poses you linked to. (Your model’s beautiful by the way – clean geometry, dynamic poses IMO.) Could you say a bit on how you go about rigging shoulders?
I apologize in advance if anyone thinks I’m trying to usurp the thread – I’m not! It’s just that this kind of modeling goes hand-in-hand with rigging and I do think the discussion could benefit more than just myself.
I know. Don’t worry. The images you posted are a great help to the progress of this thread. Any costructive discussions are great things.
About your model and rigging, I think that probably you can use some constraint to obtain some automatic behaviour like lenghtwise axis rotation of biceps.
Anyway I remind you (I hope that it can be an help) that when a person rotate forward his arm also moves forward a little clavicle and scapula.
For exaple, when walking he rotates in opposite directions the hips and the clavicles/scapulas.
Moreover remember that every movement is based on this principle. If you move forward a arm you always move backward the other to balance the body weight.
Thanks for this inspirational thread, it has made me rethink things.
Here for example, a simple edge rotation fixes a bad flow I was not aware of.
I think having a real face ring around the union of arm and thorax is very necessary. These polygons will be very skewed when this part is twisted, but will have to be stretched very much as well when other arm movements happen, so they have to be as rectangular as possible.
Maybe this images don’t show very clearly how that little edge rotation improved deformation, but the cleaner flow helps for a smoother result.
Well, it is quite an extensive subject, luckily a good part is covered in this thread.
You will not get good enough results by weight painting alone, although it is an important part. You can make use of some tricks, like constrained puling bones and lattices, but the best way to control deformatins is Ipo driven sape keys. That is how my model is rigged. I have a little tutorial in the subject.
Since Z and Y rotations seem to affect each other, this works for shoulders in a limited range, not for all the range the shoulders really have. I am trying to figure out if scriptlinks or some other tricks can help overcome this limits.
Another note:
Shoulders by no means have a movement pivoting in a fixed point. Because of the complex interaction of bones and muscles, when the arm moves forward, the whole shoulder is displaced forward (or backwards or up), actually reshaping the upper thorax, so I have tried a lot of constraints to mimic this movement. Since constraints are not made to be based on the position of a child bone, the movement is jerky, looks good for posing but not good for animation. Another problem to look into.
Great step, tolobán! And your model change is inspiring me. Now I am going to understand where could be the errors in my experiment.
And the thread is going to take the right way! Exchange of knowledge to reach a common target. Not only useless discussions about what is beautyfull and what not. Or what is academical correct and what not.
Thank you.